Python: remove hardcoded 30s default timeout from JsonRpcClient.request()

[SteveSandersonMS]

Summary

Changes JsonRpcClient.request() default timeout from 30.0 to None, so requests wait indefinitely for the server to respond — matching the behavior of the Go, Node/TypeScript, and .NET SDKs.

Related

Fixes #539

[Copilot]

Pull request overview

This PR fixes a timeout inconsistency in the Python SDK by removing the hardcoded 30-second default timeout from JsonRpcClient.request(), making it wait indefinitely like the Go, Node/TypeScript, and .NET SDKs. This resolves issue #539 where long-running RPC calls such as fleet.start would fail with asyncio.TimeoutError after 30 seconds, even though the operation was still in progress.

Changes:

• Changed JsonRpcClient.request() default timeout from 30 seconds to None (wait indefinitely)
• Updated implementation to conditionally use asyncio.wait_for() only when an explicit timeout is provided
• Updated docstrings to reflect the new timeout behavior

[github-actions]

✅ Cross-SDK Consistency Review: PASSED

This PR successfully resolves a timeout inconsistency in the Python SDK by aligning its behavior with the other three SDK implementations.

Summary of Changes

The PR removes the hardcoded 30-second timeout from JsonRpcClient.request() in the Python SDK, making it wait indefinitely by default (when timeout=None).

Cross-SDK Verification

I verified the timeout behavior across all four SDK implementations:

SDK	Default RPC Request Timeout	Status
Node.js/TypeScript	✅ Indefinite (no timeout)	Consistent
Python (before PR)	❌ 30 seconds hardcoded	Inconsistent
Python (after PR)	✅ Indefinite (no timeout)	FIXED ✨
Go	✅ Indefinite (no timeout)	Consistent
.NET	✅ Indefinite (no timeout)	Consistent

Implementation Details

Python (python/copilot/jsonrpc.py):

• Before: timeout: float = 30.0 → hardcoded 30s timeout
• After: timeout: Optional[float] = None → waits indefinitely by default
• Uses asyncio.wait_for() only when timeout is explicitly provided

Node.js (nodejs/src/client.ts):

• Uses vscode-jsonrpc with no timeout configuration
• Relies on connection-level error handling

Go (go/internal/jsonrpc2/jsonrpc2.go):

• Line 230-247: Waits on responseChan indefinitely unless process exits
• No timeout mechanism in the core Request() method

.NET (dotnet/src/Client.cs):

• Line 870-897: Uses StreamJsonRpc with CancellationToken only
• No built-in timeout; relies on caller-provided cancellation tokens

Conclusion

This change brings the Python SDK into alignment with the established pattern across all other SDKs. The fix is correct and maintains API consistency. 🎉

No additional cross-SDK changes are needed.

AI generated by SDK Consistency Review Agent

[github-actions]

✅ Cross-SDK Consistency Review

This PR successfully brings the Python SDK into alignment with the timeout behavior of the other three SDKs. The change removes the hardcoded 30-second timeout and makes Python wait indefinitely by default, matching:

• Node.js/TypeScript: Uses vscode-jsonrpc's MessageConnection.sendRequest() which waits indefinitely with no default timeout
• Go: jsonrpc2.Request() uses a simple channel select with no timeout (only exits on response, process done, or stop signal)
• .NET: InvokeWithCancellationAsync() waits indefinitely unless the caller provides a cancellation token with a timeout

Additional Changes

The other changes in this PR are type casting additions (cast(ToolResult, result) and cast(UserInputResponse, result)) which are Python-specific type checking improvements and don't affect cross-language consistency.

No consistency issues identified. This PR improves consistency across the SDK implementations. ✅

AI generated by SDK Consistency Review Agent